Conference Agenda

Remote sensing, particularly satellite-based, can play a valuable role in monitoring areas prone to geohazards. The high spatial and temporal coverage provided by satellite data can be used to reconstruct past events and continuously monitor sensitive areas for potential hazards. This paper presents a range of techniques and methods that were applied for in-depth analysis and utilization of Earth observation data, with a particular emphasis on: (i) detecting mining subsidence, where a novel approach is proposed by combining an improved U-Net model and DInSAR technology. The results showed that the ECA U-Net model performed better than the U-Net (baseline) model in terms of Mean Intersection over Union (MIoU) and Intersection over Union (IoU) indicators; (ii) monitoring water conservancy and hydropower engineering. The Xiaolangdi Multipurpose dam complex was monitored using SBAS-InSAR method on Sentinel-1 time series data and four small regions with high deformation rates were identified on the slope of the reservoir bank on the north side. The dam body also showed obvious deformation with a velocity exceeding 60 mm/year; (iii); the evaluation of the potential of InSAR results to integrate monitoring and warning systems for valuable heritage and architectural preservation. The overall outcome of these methods showed that the use of Artificial Intelligence (AI) techniques in combination with InSAR data leads to more efficient analysis and interpretation, resulting in improved accuracy and prompt identification of potential hazards; and (iv) finally, this study also presents a method for detecting landslides in mountainous regions, using optical imagery. The new temporal landslide detection method is evaluated over a 7-year analysis period and unlike conventional bi-temporal change detection methods, this approach does not depend on any prior-knowledge and can potentially detect landslides over extended periods of time such as decades.

This project consists of two sub-projects, one is focused on using an InSAR method in conjunction with Sentinel-1 satellites data to study the seismic-related studies, including interseismic deformation monitoring for strain accumulation on big faults, regional-scale deformation detection for geodynamic studies, major earthquake deformation measurements for rupture inversions, and postseismic deformation analysis for rheology studies.

Another is aimed at detecting and comprehending seismic anomalies from electromagnetic data observed by Swarm satellites present ongoing challenges within the fields of earthquake study and electromagnetism, it involves application of Long Short-Term Memory (LSTM) networks for managing time series electromagnetic data, alongside TIMEGAN (generative adversarial network), a specialized generative model for time-series generation, to identify seismic anomalies within the Swarm satellite datasets.

In the past several years, Sentinel-1 satellites acquired high-quality data for InSAR data processing, which greatly enhances the deformation monitoring capabilities over major tectonic units in China. To overcome the big data processing overburden and also obtain high-precisions as low as 1~2 mm/y in both horizontal and vertical directions for tectonic studies, we developed parallel computation systems for this purpose to utilize the Sentienl-1 catalog data as much as possible on point scatterer basis. Moreover, with the fast advancement of artificial intelligence (AI) and machine learning algorithms developed in recent years, the project has expected to integrate them into data processing systems to further improve deformation detection precisions.

On the other hand, LSTMs demonstrate commendable predictive capabilities across numerous cases, the anomaly detection using LSTM yields a notable number of false positives. Conversely, TIMEGAN models encounter difficulties in generating synthetic data, often resulting in non-informative or repetitive values. These findings underscore both the promise and hurdles associated with the application of deep learning techniques to electromagnetic data gathered by satellites.

Despite the existence of challenges from two respective sides above, the integration of the outputs poses even more challenges, which remains an unexplored avenue, offering significant potential for future research endeavors. This report will present the progressing results from both sides and an attempt at integrating tectonic background information derived from EU Sentinel 1 data, in particular possible impact of the tectonic background in detecting seismic anomalies, emphasizing the potential of deep learning methodologies in uncovering seismic precursors from satellite data. Nevertheless, these findings underscore the imperative need for further refinement and continued research in this field to enhance the efficacy and reliability of deformation monitoring and seismic anomaly detection and understanding.